mobile icon

Exploring the Effect of Voice Variation in Human – Agent Interaction

WorkgroupEveryday Media
FundingIWM budget resources
Project description

The utilization of AI-powered voice assistants (VAs) has recently become more prevalent as an emerging digital technology that aids users in their daily activities, leading also to the emergence of various forms of communication with users beyond simple commands. Therefore, it is crucial to understand how people perceive and relate to disembodied, intelligent and voice-assisted technologies in communication. Given that a significant part of communication with these virtual actors revolves around information, factors such as the credibility and perceived intelligence of the source have become more important.

This project primarily aims to understand what kind of relationship people have with voice assistants and how this interacts with other factors such as the voice of the assistant. It for example explores whether and when the female persona (e.g., voice cue) of voice assistants, which is often preferred by companies, triggers gender stereotypes that exist in society. In order to look at this more closely, it is aimed to look at the effects of different social roles (e.g., servant vs. companion) and speech styles of the voice assistant (e.g., affiliative vs. assertive) on participants' credibility judgments and perceived level of intelligence. In a next step, the plan is to examine how participants perceive the intelligence and credibility of voice assistants based on the type of information they provide.

With the increasing use of voice assistants, it is important to understand the dynamics that may influence the nature of this interaction. It is expected that the findings of these experiments will shed light on how to improve users' experiences in communicating with voice assistants. It is also hoped to contribute to the ongoing discourse on the effects of gender bias in AI technology in general.